Update to a program for saving a model fit as a dataset



20


Stata Technical Bulletin


STB-58


Ircomb requests LR tests of whether dependent categories can be combined. This option uses constrained estimation and overwrites
constraint 999 if it is already defined.

smhsiao requests Small-Hsiao tests of the IIA assumption.

set Vaarlist [ varlist...] ) specifies that a set of variables is to be considered together for the LR test or Wald test. The
backslash is used to specify multiple sets of variables. This option is particularly useful when a categorical independent
variable is entered as a set of dummy variables.

all requests that all available tests should be performed.

base also conducts an IIA test omitting the base category of the original mlogit estimation. This is done by reestimating the
model using the largest remaining category as the base category, although the original estimates are restored to memory
afterward.

Utility procedures

mlogtest uses several utility ado-files that are also used in other programs by the authors. In this section we briefly
describe these ado-files.

_perhs.ado returns the number of right-hand-side variables and their names for regression models.

_pecats.ado returns the names and values of the categories for models with ordinal, nominal, or binary outcomes. For mlogit
it indicates the value of the reference category.

Example

The data for this example are from the 1993 and 1994 General Social Survey. The nominal variable (kidvalue) is the
respondent’s choice of which of the following is most important for a child to learn to prepare him or her for life: “to obey”
(kidvalue = 1), “to think for himself or herself” (kidvalue = 2), “to work hard” (kidvalue = 3), or “to help others when
they need help” (kidvalue = 4). The fifth option, “to be popular”, was excluded because it was very rarely chosen. The
independent variables are respondent’s sex (female), race (black and othrrace, with the reference category being white),
education (degree), and whether the respondent has any children of her or his own (anykids). We begin by estimating the
MNLM.

. mlogit kidvalue female black othrrace degree anykids, nolog

Multinomial regression

Log likelihood = -3396.3518

Number of obs =
LR chi2(15)

Prob > chi2     =

Pseudo R2       =

2978

300.14

0.0000

0.0423

kidvalue

I        Coef.

Std. Err.

z

P>z

[957. Conf.

Interval]

obey

I

female

I -.2605371

.1048637

-2.485

0.013

-.4660662

-.0550079

black

I .3297048

.1452035

2.271

0.023

.0451112

.6142984

othrrace

I .5711209

.2872073

1.989

0.047

.0082049

1.134037

degree

I -.7040498

.0577797

-12.185

0.000

-.817296

-.5908037

anykids

I -.0401693

.1202552

-0.334

0.738

-.2758652

.19ББ26Б

_cons

I -.0847716

.1376452

-0.616

0.538

-.3545513

.1860081

Workhard

I

female

I -.4657661

.1104007

-4.219

0.000

-.6821476

-.2493846

black

I .1975939

.1714529

1.152

0.249

-.1384475

.Б3363Б4

othrrace

I 1.621659

.2233146

7.262

0.000

1.183971

2.0Б9348

degree

I -.1824923

.0479872

-3.803

0.000

-.2765455

-.0884391

anykids

I .0052844

.1243124

0.043

0.966

-.2383635

.2489323

_cons

I -.8719322

.1472885

-5.920

0.000

-1.160612

-.6832621

helpoth

I

female

I -.3530656

.1165728

-3.029

0.002

-.5815441

-.1246871

black

I -.1156104

.1892914

-0.611

0.541

-.4866148

.266394

othrrace

I .8759096

.2791998

3.137

0.002

.328688

1.423131

degree

I -.3875589

.0549027

-7.059

0.000

-.4951661

-.2799617

anykids

I -.1913028

.1286881

-1.487

0.137

-.4435269

.0609214

_cons

I -.5615834

.1493388

-3.760

0.000

-.8542821

-.2688846

(Outcome

kidvalue==thnkself is the

comparison group)

In the following examples, we use a series of mlogtest commands to estimate several tests. Alternatively, we could have



More intriguing information

1. AN EMPIRICAL INVESTIGATION OF THE PRODUCTION EFFECTS OF ADOPTING GM SEED TECHNOLOGY: THE CASE OF FARMERS IN ARGENTINA
2. The name is absent
3. L'organisation en réseau comme forme « indéterminée »
4. Effects of a Sport Education Intervention on Students’ Motivational Responses in Physical Education
5. Should Local Public Employment Services be Merged with the Local Social Benefit Administrations?
6. Errors in recorded security prices and the turn-of-the year effect
7. sycnoιogιcaι spaces
8. The name is absent
9. The name is absent
10. On the estimation of hospital cost: the approach